Influence of protein aggregates, extracellular vesicles, and lipoprotein fusion on ionizable lipid nanoparticles protein corona analysis

Abstract

Since 2018, ionizable lipid nanoparticles (LNPs) have revolutionized nucleic acid therapeutics. However, achieving potent extrahepatic delivery remains a formidable challenge, primarily due to rapid hepatic uptake driven by apolipoprotein adsorption. While analyzing the LNP protein corona is essential for engineering organ-specific tropism, these soft materials present unique analytical hurdles. Co-isolation of blood-borne contaminants, such as extracellular vesicles and lipoproteins, often masks the true corona composition. This perspective examines the critical need for refined proteomic strategies to distinguish genuine corona proteins from impurities. We propose tailored investigative approaches, suggesting the LNP protein corona significantly differs from the rigid shells observed on inorganic nanoparticles.

Engineering organ-specific LNPs requires a deep understanding of the protein corona, yet its fluid nature makes analysis difficult. This perspective identifies how blood contaminants mask true corona composition and advocates for refined strategies to study these unique lipid-protein interfaces.

Importance of ionizable lipid nanoparticles in nanomedicine

Genetic medicines hold unprecedented therapeutic potential, and nucleic acid-based therapeutics in particular have emerged as versatile and rapidly advancing platforms^1,2. Messenger ribonucleic acid (mRNA) has already demonstrated clinical impact through FDA-approved vaccines developed against COVID-19, validating both the feasibility and scalability of mRNA production and delivery at a global scale³. Beyond vaccines, mRNA therapeutics are being developed for cancer immunotherapy, protein replacement in genetic disorders, and regenerative applications such as tissue repair^4–7. Compared to traditional viral-based nucleic acid therapeutics, mRNA offers several inherent advantages: transient protein expression without permanent genome modification, natural biodegradability, and cost-effective, scalable manufacturing processes².

Despite these advantages, mRNA faces significant delivery barriers in vivo. For mRNA to achieve therapeutic potential, it must be transported into the cytoplasm, where ribosomes translate it into functional proteins^8,9. However, within biological fluids, single-stranded mRNA molecules are fragile and prone to rapid enzymatic degradation¹⁰. Due to its large size ( ~300–1600 kDa) and negative charge, mRNA is also unable to cross the negatively charged cell membrane³. These challenges necessitate delivery systems to protect mRNA, enable its cellular uptake, and ensure protein expression at the intended site of delivery.

Currently, ionizable lipid nanoparticles (LNPs) are the only FDA-approved delivery system for mRNA¹¹. LNPs evolved from early liposome research in the 1960s, through decades of refinement in lipid chemistry and formulation procedures^12–14. LNPs are typically composed of four components, as seen in Fig. 1: ionizable lipid, helper lipid, cholesterol, and a polymer-lipid that typically incorporates polyethylene glycol (PEG). Ionizable lipids are central for mRNA encapsulation and endosomal escape. They are designed to impart a neutral charge to the LNP surface at physiological pH to reduce toxicity and a positive charge in acidic endosomal environments. This pH-sensitive protonation enables the disruption of the endosomal membrane and the release of mRNA into the cytoplasm^15,16. Helper lipids, such as dioleoyl phosphatidylethanolamine (DOPE) and distearoylphosphatidylcholine (DSPC), stabilize the structure, provide blood compatibility, and increase the mRNA encapsulation efficiency of the LNP. Cholesterol, analogous to its role in biological membranes, provides rigidity and membrane integrity^3,17. Finally, the PEG-lipid reduces nonspecific interactions with the nanoparticle surface, thereby reducing LNP aggregation and extending circulation time in vivo¹⁸. When combined, these lipids self-assemble into spherical particles ( ~ 100 nm in diameter), with hydrophilic heads facing outward and hydrophobic tails inward, thereby encapsulating the aqueous RNA cargo.

Fig. 1.

LNPs typically comprise 4 components, each of which plays a unique role in protein corona formation, which ultimately impacts organ tropism, immune response, and cellular uptake pathways.

The performance of LNPs depends heavily on their chemical composition and the ratio of their aforementioned components. The initial LNP formulations identified in the early to mid-2000s used cationic lipids, which efficiently encapsulated RNA but also caused toxicity and inflammation. To reduce these adverse effects, cationic lipids were replaced by ionizable lipids (e.g. MC3, SM-102, or ALC-0315), thus enabling intracellular delivery with minimized inflammation and complement activation¹⁴. Additionally, the influence of other LNP components, namely helper lipids, have been studied to identify increasingly potent formulations with decreasingly immunogenic effects¹⁴. The optimization of these chemistries has been central to the success of the Pfizer/BioNTech and Moderna COVID-19 vaccines, each of which employed the helper lipid DSPC with distinct ionizable lipids: ALC-0315 and SM-102, respectively.

While initial clinical successes of LNPs have been realized for hereditary liver disease and vaccine applications³, one forefront of the RNA nanomedicine field is achieving delivery to additional organ targets to expand therapeutic indications. To accomplish this, efforts have been underway to shift LNP tropism away from the liver and into alternative tissues, and many of these efforts focus on the modification of nanoparticle chemical design. For example, it has been shown that incorporation of positively charged helper lipids such as 1,2-dioleoyl-3-trimethylammoniumpropane (DOTAP) into LNPs delivered intravenously biases delivery away from the liver and towards the lungs²⁰. Analogously, incorporating DOTAP into LNPs for intraperitoneal delivery enables substantial shifts in tropism to the pancreas²¹. Adjustments in PEG-lipid chemistry can also influence organ tropism; for example, shorter PEG chains may promote uptake in the spleen, while longer chains favor extended circulation and liver accumulation²². Beyond lipid chemistry, the attachment of targeting ligands, such as sugars or peptides, can mediate receptor- and cell-specific uptake in tissues, including the immune system and tumor microenvironments²³. However, while these modified LNP formulations have biased mRNA delivery towards extrahepatic organs, these formulations still mediate significant hepatic delivery, and LNP biodistribution beyond the liver remains a formidable limitation.

Thus, the mechanism of LNP chemistry-dependent shifts in tropism has been of increasing interest, with a leading explanation involving the “protein corona”, which is a layer of biomolecules, mostly proteins, that adsorb to the LNP surface upon introduction into biological fluids^24–28. The protein corona influences particle identity, blood circulation time, pharmacokinetics, cellular uptake, and immune recognition^29–34. The composition of the protein corona is governed mainly by three major factors: the physicochemical properties of the nanoparticles (NPs), the specific characteristics of the biological fluid, and the environmental conditions of the exposure of the NP to the fluid, such as incubation time, temperature, and fluid dynamics; these factors have all been heavily investigated to tune nanoparticles pharmacokinetics in vivo³⁵.

In many cases, it is the protein corona, rather than the bare nanoparticle surface, that dictates biological outcomes³⁶. For LNPs, this was first observed when upon systemic administration, they preferentially accumulated in the liver. It was demonstrated that the liver-tropic LNPs disproportionally adsorbed apolipoprotein E (ApoE) following introduction to blood³⁷. Hepatocytes highly express the low-density lipoprotein receptor (LDLR), which has high affinity for ApoE; thus, ApoE-adsorbed LNPs are preferentially endocytosed within liver cells^37,38. Additionally, LNPs incorporating a permanently-charged helper lipid excipient have been shown to shift tropism away from the liver following intravenous administration; positively charged lipids, such as the aforementioned DOTAP, promote delivery to the lungs, whereas negatively charged lipids, such as 1,2-dioleoyl-sn-glycero-3-phosphate (18PA), promote delivery to the spleen¹⁹. More recent work found that these tropic shifts occur due to the five-component LNPs attracting unique protein coronas that are not enriched in ApoE, but rather enriched in vitronectin (Vtn) or β2-glycoprotein I (β2-GPI), which promotes LNP endocytosis through receptor-mediated interactions with pulmonary endothelial cells and splenic macrophages, respectively³⁹. These results demonstrate the power of correlating LNP chemistry with protein corona composition; it is essential for the rational design of extrahepatic, organ-specific mRNA delivery systems.

Despite these insights, a major challenge in the field remains: discerning whether the variations in biological outcomes are driven primarily by differences in the protein coronas or by differences in the inherent physicochemical properties of the LNPs, or a combination of both factors. This difficulty arises because analyzing the protein coronas on LNPs, which are classified as soft nanoparticles, is inherently more complex than on hard nanoparticles, such as gold or polystyrene, owing to the challenges in removing protein aggregate contaminants from serum-incubated LNPs and isolating only the nanoparticle corona proteins^40,41. Accordingly, precise characterization of a contaminant-free, well-defined protein corona is crucial for optimizing extrahepatic LNP design³⁵.

Achieving these goals will require not only technological innovation but also rigorous methodological approaches to studying LNP-biological interactions, beginning with accurate characterization of the protein corona itself. In this perspective, we discuss the key challenges involved in isolating, imaging, quantifying, and interpreting the protein corona. We also propose future strategies to address these hurdles and improve our understanding of the protein corona on LNP surfaces and other soft nanoparticles (e.g., liposomes). Notably, we suggest that the corona on LNPs may differ significantly from the traditional shell-like protein layer observed on hard nanoparticles (e.g., metal, metal oxide, and plastics), highlighting the need for tailored investigative approaches.

Challenges in the isolation of LNP protein coronas

Accurate protein corona characterization relies upon the ability to separate corona-coated NPs from molecular contaminants, such as extracellular vesicles (EVs) and protein aggregates. For nanoparticles made of hard materials, protein corona formation and characterization methods are well-established because the corona-coated nanoparticles can be easily separated from proteinaceous contaminants in biological fluids⁴², typically by centrifugation. Unfortunately, these methods may not reliably separate LNPs from EVs, such as exosomes, due to their similar physicochemical properties^43,44, including lipid composition, density, and size⁴⁵. As a result, samples of protein corona-covered LNPs can become contaminated with proteins present within EV membranes. Moreover, the similar surface properties of EVs and LNPs drive the irreversible aggregation, fusion, and/or combination of EV and LNP protein coronas⁴⁶. These artifacts can significantly compromise the interpretation of proteomic data through mass spectrometry, which enables identification of the protein corona’s compositional profile, including the types of proteins, their post-translational modifications, and their relative abundances^{24–27,30,41,47–49}.

Conventional size- or density-based fractionation methods, including size-exclusion chromatography, often isolate LNPs, unbound endogenous proteins, and EVs in the same eluate fractions. This makes high-purity, high-throughput recovery of protein-coated LNPs difficult, as further receptor-based purification steps are necessary to ensure that the unbound protein and EV contaminants are removed. While many LNPs can be formulated in a high-throughput manner⁵⁰ for in vitro or in vivo validation, the current process of isolating LNP-specific protein coronas is too arduous to complete for all formulations and greatly limits the utility of protein corona engineering for extrahepatic LNP delivery³⁵.

For LNPs conjugated to active targeting moieties, such as specific antibodies, affinity-based pull-down isolation methods have emerged as a promising approach to separate protein-coated LNPs from a biological fluid, however this methodology still risks the co-isolation of endogenous EVs due to the potential for the targeting antibodies to dissociate from the LNPs and embed within the contaminating EVs^51,52.

Visualization of the protein corona and its molecular contaminants

Microscopy is often used to visualize the protein corona and confirms its presence after NP incubation within a biological fluid. Imaging the intact and hydrated protein corona allows for the identification of impurities such as EVs and protein aggregates. However, only a few techniques are capable of directly visualizing the precise morphology of the protein corona at the nanoscale^{35,40,41,53,54}.

One such technique is cryo-electron microscopy, which, when combined with cryo-electron tomography, enables direct visualization of the three-dimensional architecture of a nanoparticle with its protein corona in their natural, hydrated state (Fig. 2). Unlike indirect methods, such as dynamic light scattering, field flow fractionation, or differential centrifugation, this approach offers direct, nanoscale insights into the arrangement of proteins on the nanoparticle surface (Fig. 2b) in contrast to contaminants in the background of the images (Fig. 2e). Additionally, from the images in Fig. 2, it is notable that a hard protein corona is a tightly bound, stable layer of proteins on the nanoparticle surface, whereas a soft protein corona consists of loosely associated, dynamically exchanging proteins.

Fig. 2. Cryo-TEM and 3D reconstruction images of the protein corona on nanoparticles.

Polystyrene nanoparticles and gold fiducial markers were incubated with human plasma prior to cryo-TEM imaging. Nanoparticles were imaged (a) before and (b) after hard, shell-like protein corona formation, with the gold markers appearing as black dots. c Snapshot of a movie generated from the 3D volume of the protein corona. The full view of the 3D reconstruction can be seen in Supplementary Movie 1. d High magnification views of the protein corona on the surface of polystyrene nanoparticles, revealing contamination (black arrows) as large clusters within the corona layer. e Depiction of a soft corona shell at the surface of polystyrene nanoparticles with (f) an electron tomogram slice of the corona layer, superimposed with a colored 3D reconstruction. Figure panels and the supplementary movie are reproduced with permission^53,54.

While cryo-imaging has been used extensively to visualize protein coronas among polymeric and metal nanoparticles, to the authors’ knowledge, only one recent study has directly visualized the LNP protein corona using cryo-TEM (Fig. 3)⁴³. The researchers implemented an innovative isolation approach using magnetic nanoparticles to isolate the LNPs after their exposure to human serum, aiming to minimize the contamination of the LNP protein corona sample by protein aggregates and EVs. Although this isolation method reduced contamination (Fig. 3a) compared to traditional, non-magnetic column separation (Fig. 3b), it is interesting to note that a hazy protein corona is not clearly visible around the surface of the LNPs in either separation method. This result contrasts with  the characteristic protein coronas observed on the surface of hard polystyrene nanoparticles, as illustrated in Fig. 2^53,54.

Fig. 3. Cryo-TEM images illustrating EVs and LNPs following interaction with biological fluids.

Cryo-TEM images of LNPs after incubation with human serum and following separation from excess serum using (a) magnetic column separation and (b) traditional, non-magnetic column separation. Serum-derived vesicles (white arrows) are clearly visible in both samples; the molecular morphology of the protein corona is not clearly visible around LNPs. Scale bars represent 50 nm, and figure panels are reproduced with permission⁴³.

We recently validated this finding, where we observed that no conventional protein corona forms on the surfaces of various LNPs containing DOPE, DSPC, DOPS, and DOTAP (Fig. 4a-d)⁵⁵. As a positive control for typical protein corona formation on hard nanoparticles, we incubated polystyrene nanoparticles with human plasma, isolated the corona-coated particles, and imaged them using cryo-TEM (Fig. 4e). As expected, these nanoparticles exhibited a clear, uniform protein corona (green arrows in Fig. 4e)⁵⁵. Interestingly, we also detected some remaining EVs in the same fields, which, similar to the LNPs, showed little to no corona-like surface coverage (red arrows in Fig. 4e).

Fig. 4. Cryo-TEM images of various types of LNPs after interactions with plasma proteins do not indicate conventional protein corona formation.

Analysis of ionizable lipid nanoparticles (LNPs) containing various helper phospholipids: (a) DOPE, (b) DOPS, (c) DOTAP, and (d) DSPC. The LNPs were formulated using the ionizable lipid 306Oi10, a helper lipid (DOPE, DOTAP, DSPC, or DOPS), cholesterol, and C14-PEG2000 at a molar ratio of 35:16:46.5:2.5, respectively, with a lipidoid-to-mRNA mass ratio of 10. Characterization via DLS (Dynamic Light Scattering) and TNS (2-(p-toluidino)−6-naphthalenesulfonic acid) assay revealed that all formulations maintained an average diameter between 85 and 103 nm and a narrow pK_a range of 6.2–6.4. While all LNPs exhibited a slightly negative surface charge (zeta potential ranging from −4.02 to −10.3 mV), mRNA entrapment varied significantly by helper lipid; DOPE and DOTAP facilitated high encapsulation ( > 82%), whereas DSPC and DOPS resulted in lower efficiencies of 36.8% and 24.7%, respectively. e The analysis of polystyrene nanoparticles (bottom row) of two different sizes (100 nm and 200 nm) demonstrated the presence of a characteristic fuzzy protein corona (indicated by green arrows), whereas extracellular vesicles (EVs) within the sample lacked such a structure (red arrows). These findings highlight the unexpectedly minimal or absent protein corona on LNP surfaces, contrasting sharply with the well-documented corona formation observed in polystyrene nanoparticles in existing literature. Figure panels are reproduced with permission⁵⁵.

Consistently, cryo-TEM images revealed a continuous protein layer on the surface of the polystyrene nanoparticles, while various LNPs and EV surfaces remained largely devoid of comparable protein adsorption. This is notable because EVs are more structurally similar to LNPs than polystyrene nanoparticles, prompting us to caution against over-interpreting corona signatures on LNPs. We suggest that there are mechanistic differences in corona formation between synthetic nanoparticles and biologically inspired nanoparticles, such as LNPs, due to their greater resemblance to biological vesicles, such as EVs.

Biological lipid vesicles lack a conventional shell-like protein corona

Consistent with this finding, the cryo-TEM images presented in many studies rarely reveal a discernible protein corona deposited on the surface of EVs of various shapes and sizes (Fig. 5a-c); only one study to the authors’ knowledge directly visualizes distinct characteristics of a protein corona observable on subtypes of small porcine EVs (Fig. 5d, e)^56–60. While there is clear non-membranous density evident in these images, it is difficult to discern if the vesicles are coated with protein, as the ‘halo’ that engulfs the EV is highly regular, unlike most protein corona surfaces observed in cryo-EM images of polystyrene nanoparticles.

Fig. 5. The minimal formation of a distinct protein corona structure on imaged EVs.

Cryo-TEM images displaying a variety of EVs with different sizes, structures, and contents. Clusters of proteins and other biomolecules are visible in the background of the images, but the EVs lack discernible protein corona formation on their surfaces. a Images of single vesicles (white arrow), double vesicles (red arrow), multilayer vesicles (blue arrow), and vesicles containing electron-dense cargo in the lumen (yellow arrow). b Images of double-membrane vesicles (green arrow), vesicles with electron-dense cargo in the lumen (yellow arrow), vesicles resembling sacks (white arrows), and distinctly elongated vesicles (black arrow). c Images of multilayer vesicles (blue arrows). d, e Representative cryo-EM images of subtypes of small porcine EVs showing the distinct protein corona structure surrounding their outer membrane (yellow arrows). Figures are reproduced with permission from refs. ^56,58.

Some studies have inferred the formation of a protein corona on EVs⁶¹, based on data from dynamic light scattering and gel electrophoresis after EVs are isolated via ultracentrifugation, however these coronas were not confirmed with imaging. Additionally, while ultracentrifugation is a common method of EV isolation from biological fluids⁶¹, this method carries a substantial risk of sample contamination with lipoproteins and other biomolecules that are co-isolated and are fused with EVs under these conditions; such contaminants can distort mass spectrometry-based characterization of the protein corona and bias interpretations of protein and biomolecule interactions with the surface of EVs⁴⁰.

Given the structural parallels between EVs and LNPs, we expect protein coronas on LNPs to differ from those on other nanoparticles (Fig. 6). Specifically, for unmodified LNPs, we hypothesize that the dense, multilayer protein shell characteristic of hard nanoparticle coronas does not form on LNPs. Instead, protein association with LNPs may be sparse, patchy, or structurally integrated into the lipid bilayer in ways that are functionally significant but not resolvable as a distinct surface layer by cryo-TEM. Much like EVs, LNPs self-assemble as spherical bilayers in a manner that exposes their hydrophilic and often charged functionalities to the aqueous environment, effectively sequestering their hydrophobic tails from the polar milieu. To a large extent, cytosolic protein folding energetics also drives non-polar residues to ‘hide’ in the interior, leaving more polar groups on the surface. At first glance, this arrangement would seem favorable for the interaction of proteins and LNPs, yet polar interactions between proteins and the functional groups on the surface of LNPs are thermoneutral at best, as the aqueous buffer’s polar components are available in much higher molar excess over LNPs and plasma proteins.

Fig. 6. Association of proteins with LNPs.

a Polarity-driven interactions of soluble proteins with the surface of a nascent LNP are expected to be weak. As opposed to the entropically favored hydrophobic interactions that favor strong binding in aqueous media (energetically favorable due to the release of ordered water molecules), electrostatic interactions involving one-to-one exchange of anions or cations offer little thermodynamic advantage. Moreover, the associated dehydration and rehydration of ionic or polar functionalities are unlikely to result in a significant increase in entropy. The hydration shell surrounding proteins (depicted as the orange halo in the illustration) is reduced upon its binding with the surface of the LNP. b In contrast to the latter discussion, post-translationally modified lipidated proteins are more likely to associate with LNPs through insertion of the lipid within the lipid bilayer. This entropically-driven hydrophobic effect ‘hides’ the difficulty to solvate hydrocarbon, resulting in an energetically favorable interaction.

Here, we assume a dynamic equilibrium is in play such that electrostatically adsorbed proteins can equilibrate with their dissociated state and be solvated in the aqueous media, irrespective of whether the protein-LNP interactions occurred prior to or after the incubation in biological fluid. Energy compensation would be expected when the plasma proteins’ ionic and polar interactions with buffer salts and water are replaced by similar interactions with the LNPs' surfaces, while entropic costs would most likely disfavor the overall process. Thus, for traditional LNPs without surface modifications, formation of a dense, uniform protein shell akin to that observed on rigid nanoparticles should be thermodynamically disfavored. This does not preclude selective protein association through alternative mechanisms, including hydrophobic insertion of lipidated proteins, lipoprotein fusion, or low-density adsorption of high-affinity species, all of which have been demonstrated experimentally through proteomic and functional studies of LNP corona proteins.

Additionally, the lack of traditional protein corona formation on EV surfaces might stem from specific components of EV membranes that may discourage protein adhesion. As such, systematic studies are needed to investigate how precise LNP lipid composition and stoichiometry modulate protein interactions and corona formation; the results of these studies would have direct implications for the design and synthesis of LNPs capable of endogenous, protein-corona-mediated organ targeting. Insights from native lipid membranes, comprised of bulk, annular, and nonannular lipids⁶², should be used to inform models of protein-lipid interactions and deepen our field’s understanding of the formation and nature of protein coronas on LNPs and EVs.

Current studies⁶² on protein-lipid interactions have suggested various proteins, including G-protein coupled receptors (GPCRs), tetraspanins, channels (such as Aquaporin Z and AmtB), and transporters (SERT and Na⁺/H⁺ antiporters), may bind selectively to LNPs, although definitive experimental proof has yet to be reported. Future research will likely shed light on the structural determinants of these interactions and their roles in LNP functionality, such as protein corona formation upon introduction to protein-rich biological fluids. It is important to highlight, however, that bulk protein interactions with nanoparticles, which constitute the traditional protein corona, are unlike specific or membrane-associated interactions of proteins that tend to embed themselves within lipid bilayers and may embed themselves within structurally similar LNPs. Lipid-protein interactions are critically implicated in cellular metabolism, signaling, trafficking, and membrane organization, and recent advances in analytical, biochemical, and computational technologies have revealed how both specific lipid-binding motifs and the collective behavior of lipids modulate protein structure and function⁶².

It is also important to note that many of the proteins that are associated with LNPs might not be the result of conventional ‘protein corona’ formation, but rather specific interactions that drive the assembly of LNP-protein aggregates/fusion. Lipidated proteins, such as those bearing S-palmitoyl, N-myristoyl, or isoprenyl (farnesyl/geranylgeranyl) groups, associate with membranes by hydrophobic insertion of their lipid moieties into the bilayer. Palmitoylation, for example, is a reversible modification that can turn on or off, controlling how proteins move and settle within cell surfaces, especially those involved in signaling^63,64. In contrast, prenylation adds a more permanent, hydrophobic anchor that embeds deeply into the membrane⁶⁵. Direct biophysical studies of these lipid groups show spontaneous insertion into lipid bilayers and liposomes, rapid intermembrane exchange on the scale of seconds, and domain preferences that depend on lipid composition (ordered vs. disordered phases)^65–67.

Because LNPs present an outer lipid leaflet enriched in phospholipids, and depending on their formulation, cholesterol, and PEG-lipids, the same hydrophobic insertion principles allow lipidated proteins and peptides to embed or fuse into LNP surfaces or to associate via already fused lipid-binding proteins. Indeed, ApoE adsorbs to circulating LNPs⁶⁸, reorganizes their lipid architecture, and mediates hepatocyte uptake via LDLR-family receptors—a well-documented pathway for siRNA/mRNA LNP delivery to LDLR-rich hepatocytes. LNP composition and helper lipids modulate the amount and identity of adsorbed proteins, influencing cellular uptake and biodistribution; this supports the broader premise that lipidated motifs can engage LNP membranes directly or via lipid-binding corona proteins in biological fluids⁶⁹. Engagement with lipoproteins can endow EVs with a distinct biological identity, reflected in altered trafficking pathways and recipient-cell responses such as cytokine secretion^70,71. Analogously, lipoprotein-driven identity changes may plausibly extend to LNPs that share physicochemical characteristics with EVs.

A recent report presents a rapid lipoprotein association fluorometry (LAF) assay that enables semi-quantitative prediction of lipoprotein binding to EVs, small molecules, and synthetic nanoparticles. In tests with synthetic lipid nanoparticles (with and without PEG-lipid), the assay revealed detectable binding for non-PEGylated formulations to very low density lipoproteins (VLDL)⁴⁶. Modern LNP formulations use PEG to promote self-assembly, stabilize particles, and reduce surface interactions that would otherwise promote immune cell uptake and clearance. Maximization of in vivo hepatic delivery requires the PEG layer to be shed after intravenous administration. Therefore, the PEG-lipid fraction is minimized, and the lipid tails are kept short⁷². Interestingly, EVs isolated from cancerous and non-cancerous human NK cells showed distinct binding to VLDLs, underscoring the need to account for background contamination when analyzing protein-binding patterns. Overall, this study shows that lipid-based particles broadly can exhibit selective lipoprotein engagement and further can traffic and distribute bound proteins across different lipid-composed particles.

Our recent cryo-TEM analysis of various LNPs following incubation with plasma proteins revealed fusion behaviors similar to those frequently observed in EVs⁴⁶. Specifically, we detected substantial fusion events involving different lipoproteins in human plasma attaching to or merging with the LNPs (black arrows in Fig. 7). These structural findings provide a mechanistic bridge to recent functional studies of LNP protein coronas. Previous research has established that plasma proteins—particularly apolipoproteins like ApoE—partition onto LNP surfaces upon systemic exposure, profoundly influencing cellular uptake, intracellular trafficking, and mRNA transfection efficiency³⁷. For instance, Voke et al.⁷³ recently used continuous density gradient ultracentrifugation to identify a small set of proteins that reproducibly enrich on LNPs, including vitronectin, C-reactive protein, and alpha-2-macroglobulin. Functional studies demonstrated that these proteins modulate delivery outcomes in unexpected ways: certain corona proteins enhanced cellular entry while paradoxically inhibiting mRNA expression, likely by diverting particles toward lysosomal degradation. Notably, the number of enriched proteins identified by Voke et al. was small relative to prior studies, consistent with the view that protein association with LNPs is selective and sparse rather than forming a dense, uniform shell. These functional data confirm that proteins do associate with LNPs in biologically meaningful ways, even if the structural character of that association differs from the classic, hard shell-like protein corona observed on rigid nanoparticles.

Fig. 7. Cryo-TEM images illustrating lipoprotein association and fusion with the surfaces of EVs and LNPs.

a Representative cryo-TEM images of EVs isolated from biological fluids. Although no conventional protein corona is observed, these images show binding and fusion events between EVs and lipoproteins, indicated by black arrowheads (scale bar = 50 nm; figures adapted with permission from ref. ⁴⁶). b Representative cryo-TEM images of various types of LNPs, demonstrating that instead of forming a distinct, classical protein corona shell, lipoproteins directly associate with and fuse into the LNPs (black arrowheads), indicating a fusion-driven interaction rather than traditional corona formation. Specimen Preparation: To ensure structural preservation, samples were vitrified using a Thermo Fisher Scientific Vitrobot Mark IV system. This automated plunge-freezing method maintains constant environmental parameters—including temperature, relative humidity, and standardized blotting force—to facilitate rapid cryo-fixation. This process ensures that LNPs and EVs are captured in their native, hydrated state without the introduction of ice-crystal artifacts or structural damage. Figure panel b is reproduced with permission⁵⁵.

The findings presented in Fig. 7 fundamentally redefine the structural landscape of LNP-protein interactions, prompting a shift from the “adsorbed shell” model to a paradigm dominated by lipoprotein fusion, association, and transient “bleb” interactions⁵⁵.

This paradigm shift has three fundamental implications for biodistribution and design of LNPs: 1) Targeted delivery, in which achieving cell-specific targeting requires acknowledging lipoprotein-mediated uptake (e.g., via the ApoE/LDLR axis) as the primary biological driver. Consequently, engineering specific lipoprotein “coronae” becomes a central design goal, achievable by modulating lipid molar ratios or incorporating ApoE-mimetic ligands. 2) Rational design, in which rather than empirically balancing PEGylation against uptake, formulations can now be rationally engineered to govern specific lipoprotein fusion events through the coordinated optimization of PEG density, ionizable lipid chemistry, and surface charge. 3) Stability and release where the intimate nature of lipoprotein-LNP fusion directly impacts membrane integrity and cargo release kinetics. Future LNP design must leverage these “productive” fusion events to facilitate endosomal escape, necessitating the development of advanced assays to quantify these specific fusion and bleb-mediated interactions in vivo.

These findings challenge the traditional view that LNP surfaces form a uniformly structured protein corona, akin to that of hard nanoparticles. Instead, protein fusion and association processes appear to dominate the biological interface of lipid-based nanoparticles and lipoproteins. Accurate prediction of protein corona formation on LNPs requires careful consideration of their full chemical composition and the environmental conditions influencing their interactions with lipoproteins and plasma proteins. Incorporating additional components—such as lipoproteins, lipopolysaccharides, PEGylated lipids, or cholesterol—can significantly alter LNP surface properties, modulating the strength and specificity of protein interactions^11,74. Small compositional adjustments may markedly influence corona composition, thereby impacting LNP tropism and organ-specific delivery in clinical settings.

It is noteworthy that the distinction between the protein coronas of LNPs and those of more traditional polymeric or metal nanoparticles is rooted in both mechanical stability and surface chemistry. While polymeric nanoparticles are “softer” than metals, they often possess a solid or semi-solid core that supports a persistent “hard” corona—a layer of tightly bound proteins that remain associated over long timescales. In contrast, LNPs are characterized by a fluid lipid bilayer; we propose that wherever a lipid-like shell exists, the interface is governed by fusion-like interactions rather than simple surface adsorption.

While this fusion mechanism may occur on any lipid-encapsulated particle, it is most pronounced in soft LNPs where the membrane’s fluidity allows for the dynamic intercalation of apolipoproteins. These structural differences necessitate distinct isolation and characterization protocols³⁵. Standard pelleting ultracentrifugation is poorly suited to LNPs because their low density and structural fragility lead to deformation or co-sedimentation with endogenous contaminants. However, continuous density gradient ultracentrifugation, when paired with appropriate biofluid-alone controls, has recently been shown to achieve effective separation of protein-LNP complexes from native plasma particles while preserving colloidal stability⁷³. Complementary approaches such as field-flow fractionation may provide additional resolution for separating LNP subpopulations, and further methodological development in this area remains an important priority.

Ultimately, it is critical to recognize that the lipoproteins and plasma proteins capable of fusing with the LNP membrane exhibit significant variability in abundance, function, and conformation depending on the host milieu and the specific biological context⁵². Clinical factors—including the patient’s disease state, comorbidities, and demographic variables such as age, sex, and metabolic status (e.g., fasting vs. fed)—can profoundly reshape the endogenous lipoprotein profile, thereby shifting the “fusion-like” corona composition^75–77. Beyond biological variability, the choice of biofluid and experimental parameters—such as the use of serum versus plasma, the specific anticoagulant (e.g., EDTA vs. heparin), protein concentration, and the local pH or ionic strength—can fundamentally alter fusion kinetics and biomolecular exchange profiles⁷⁸. Consequently, achieving a reproducible and translatable understanding of LNP-biological interfaces requires a rigorous methodological standard. Researchers must not only transparently specify these variables but also actively control for them to ensure that the “biological identity” of the LNP is accurately characterized, facilitating more predictable outcomes in its intended therapeutic application.

Strategies for improved protein corona analysis of LNPs

To better understand the true shape and structure of the protein corona on LNPs and how it differs from conventional protein coronas on rigid nanoparticles, future studies should incorporate cryo-TEM analysis of LNPs after interactions with biological fluids. This visualization technique provides a detailed view of how proteins associate with LNP surfaces; visualizing these interactions, particularly those of lipoproteins and their fusion within LNPs, and their dependence on various LNP formulations, can enable researchers to better understand the factors driving differences in biodistribution in vivo. Changes in LNP composition, such as altering the ratio of ionizable lipid, helper phospholipid, and cholesterol⁷⁹, have been shown to significantly impact biodistribution by increasing the association of immunoproteins and decreasing apolipoproteins upon the LNP surfaces, shifting accumulation away from the liver to other organs⁸⁰. Similarly, studies have shown that simplifying LNP formulations by removing cholesterol and helper phospholipids can shift tissue tropism from the liver to the lungs⁸¹. Formulations based on ionizable cationic lipids with biodegradable ester cores, combined with permanently cationic lipids and PEG-lipids, have achieved robust pulmonary mRNA accumulation and translation, primarily in endothelial and epithelial cells⁸¹. Cryo-TEM analysis of how compositional changes influence fusion behaviors and the associated reduction in apolipoproteins can enhance our understanding of how such modifications impact LNP biodistribution and their bio-nano interface dynamics.

To address the challenges of isolating, visualizing, and composing protein coronas formed among LNPs, researchers must re-evaluate current methodologies and implement more refined strategies that effectively investigate protein corona formation and apolipoprotein fusion while addressing the prominent issue of EV contamination⁵⁵. One approach involves the removal of EVs from the biological fluid prior to incubation with LNPs^55,82. This can be accomplished with a number of established techniques, such as ultracentrifugation or magnetic-activated cell sorting (MACS)⁴³. The purified proteins can then be incubated with LNPs to analyze the corona—primarily lipoprotein fusion within the lipid membrane—without contamination from other sources.

While the kinetics of conventional protein corona formation on rigid nanoparticles are well-documented^{25–27,83–86}, typically reaching a stable “hard” corona state within a standard one-hour incubation, our understanding of the fusion-based corona on LNPs and natural lipid vesicles (EVs) remains in its infancy. Unlike the surface adsorption seen on hard materials, the LNP biological interface is likely an evolving mosaic of lipid and protein integration. Future research must, therefore, prioritize characterizing how this fusion mechanism matures over timescales ranging from minutes to several hours, and investigate the extent to which secondary proteins interact with or anchor to these newly fused domains.

Within the complex milieu of the bloodstream, LNPs do not exist in isolation; they undergo continuous biomolecular exchange with endogenous nanoparticles, including lipoproteins [HDL (high-density lipoprotein), LDL, VLDL] and EVs. This interplay is not merely competitive but involves the reciprocal transfer of lipids and the intercalation of lipoprotein components into the LNP bilayer, effectively “maturing” the particle’s identity as it circulates. Such dynamics have profound implications for characterization; a “snapshot” analysis taken at a single time point may fail to capture the transient or evolved corona composition that ultimately dictates organ tropism at the moment of cellular uptake. Consequently, achieving a translatable understanding of LNP behavior requires longitudinal studies that map the corona across multiple time points, providing a high-resolution trajectory of the LNP’s evolving biological identity in vivo.

Finally, it is crucial to recognize that certain proteins that attach to LNPs are derived from EV contamination rather than plasma adsorption. These EV-associated proteins are commonly found within small EVs (30–150 nm) and serve functional roles in vesicle formation, trafficking, and intercellular communication; however, their presence can serve as distinctive biomarkers for EV contamination within LNP formulations and signal to researchers that their LNPs require further purification for accurate proteomic analyses of their corona. To explore these potential biomarkers of contamination, we utilized publicly available datasets and compared them against the Human Protein Atlas: specifically, we compared the top 100 small EV-associated proteins identified across over 1200 studies (http://exocarta.org/sEV_top100; Supplementary Data 1) against 4294 proteins detected in human plasma via mass spectrometry (https://www.proteinatlas.org/humanproteome/blood/proteins+detected+in+ms; Supplementary Data 2).

Our comparative analysis revealed that 10 of the EV-associated proteins were notably absent from the plasma proteome, highlighting their probable specificity to EVs rather than plasma (Table 1). This indicates that the detection of these EV-enriched proteins within the protein corona of LNPs does not merely reflect plasma protein adsorption but may instead signify residual EV contamination. When these proteins appear in nanoparticle protein corona datasets, additional validation steps should determine whether they originate from genuine corona formation or from residual EV contamination. This distinction may affect the interpretation of NP-biological interactions and the assessment of therapeutic efficacy. Ultimately, incorporating such differentiation strategies will improve the reliability of LNP protein corona characterization, thereby refining LNP design for more targeted and effective therapeutics.

Table 1.

The ten most abundant proteins in small extracellular vesicles (EVs) that are not listed among the 4294 proteins detected in human plasma, as reported in the Human Protein Atlas

Gene	Protein accession number (UNIPROT)
CD63	P08962
ACTB	P60709
RAP1B	P62078
CD274	Q15116
HIST1H4A	P62805
EPRS	Q8N3J0
ATP5B	P15692
ARF1	P84095
SLC16A1	P53985
HGS	Q08430

Perspective and future vision

The future of precision nucleic acid therapeutics hinges on transitioning from empirical LNP design to a mechanistic understanding of the bio-nano interface. While ApoE-mediated hepatic tropism provides a successful template for current clinical LNPs, expanding this toolkit to non-hepatic pathologies requires a departure from traditional nanoparticle paradigms. We envision a conceptual shift: viewing LNPs not as rigid scaffolds for structured protein shells, but as dynamic assemblies that engage in fusion and integration with endogenous lipoproteins and vesicles. Recognizing that these interactions are often governed by hydrophobic insertion and lipid-phase merging, rather than simple electrostatic adsorption, is essential for the rational design of organ-specific nucleic acid delivery systems. Understanding these phenomena is crucial to optimizing LNP design, improving targeting specificity, and predicting in vivo behavior.

Realizing this potential requires developing high-resolution, contamination-aware analytical workflows. Overcoming the persistent challenges of co-isolating extracellular vesicles and protein aggregates is a prerequisite for establishing a reliable proteomic “identity” for LNPs. Future research should leverage advanced cryo-electron tomography and integrated multi-omics to map these complex association events in situ. By standardizing these refined methodologies, the community can move toward a predictive model of LNP fate, ultimately enabling the programmable delivery of genetic medicines across diverse physiological landscapes.

Author contributions

B.B., A.M.M., A.A.S., B.G., J.E.J., N.M., M.J.M., H.V., K.A.W. and M.M. contributed to conceptualization, literature review, drafting the original manuscript, and reviewing the final version. B.B., K.A.W. and M.M. led the conceptualization and supervised the manuscript preparation. All authors reviewed and approved the final version of the manuscript.

Peer review

Peer review information

Nature Communications thanks Ashish Kulkarni and the other anonymous reviewer for their contribution to the peer review of this work.

Competing interests

M.M. is co-founder and director of the Academic Parity Movement (www.paritymovement.org) (a non-profit organization dedicated to addressing academic discrimination, violence and incivility) and co-founder of Targets Tip, AlbuDerm, and XProteome Inc., and receives royalties/honoraria for his published books, plenary lectures, and licensed patents. B.B. and A.A.S. are co-founders of XProteome. The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-74240-4.